Brazilian fintech Trio built a unified payment analytics platform on ClickHouse Cloud to handle 243M+ payments in H1 2025 and 1B+ daily events. By replacing fragmented legacy systems with ClickHouse as a single source of truth, they achieved 88% storage reduction and dramatically faster query performance. Their architecture uses Redpanda for streaming, Airflow for batch ingestion, and a custom ETL layer for schema control. To handle late, duplicate, and out-of-order payment events, they implemented a sliding window pattern using refreshable materialized views and ReplacingMergeTree for deduplication, allowing recent data to stay accurate without full dataset rebuilds.
Table of contents
The need for a single source of truth #Trio’s ClickHouse-based architecture #The sliding window solution #“A generational leap” #What’s next for Trio and ClickHouse #Sort: